Acceptable Use Policy

1. Scope

This policy applies to all Refini users and to all image-only content processed through Refini, including uploaded images, reference images, AI-edited outputs, AI-generated images, prompts, text instructions, and related account activity.

Refini does not support audio, video, model uploads, dataset uploads, or non-image media processing through the public product.

2. Prohibited Content and Conduct

You may not use Refini to request, upload, generate, edit, store, distribute, or attempt to create illegal, unsafe, abusive, deceptive, or rights-infringing content.

• Child sexual abuse material, sexual content involving minors, child exploitation, grooming, or any sexualized depiction of a minor.
• Terrorist, extremist, or mass violence content, including praise, recruitment, operational support, or instructions.
• Instructions for weapons of mass destruction, serious physical harm, evasion of law enforcement, or violent wrongdoing.
• Non-consensual intimate imagery, sexual deepfakes, or sexualized images of a real person without consent.
• Pornographic, nude, sexually explicit, NSFW, or sexually suggestive content.
• Hateful, harassing, threatening, abusive, defamatory, or violent content targeting people or protected groups.
• Impersonation, deceptive identity claims, misleading AI-generated content, fraud, spam, phishing, or scams.
• Content that infringes intellectual property, privacy, publicity, or other legal rights.
• Prompts or behavior intended to bypass moderation, safety controls, rate limits, payment controls, or platform security.

3. Restricted Use

Refini is a creative image-editing and image-generation service. It must not be used as professional legal, medical, financial, identity verification, biometric identification, or public safety advice. AI outputs may be inaccurate or incomplete, and users are responsible for reviewing generated results before using them.

4. Moderation and Safety Controls

Refini uses automated and manual safety controls to reduce misuse. Uploaded images are validated for supported file type and size, hashed, moderated before storage or processing, and logged for audit evidence. Prompt-based generation and supported text instructions are also checked before AI processing.

Refini may use safety providers such as OpenAI, Replicate, and other trusted moderation or AI infrastructure providers. If moderation is unavailable or a request cannot be verified, high-risk image processing is blocked rather than allowed by default.

• Input filtering: prompts, text instructions, uploads, and reference images may be reviewed for unsafe intent or content.
• Output controls: generated or edited images may be blocked, removed, restricted, watermarked, or otherwise limited when safety signals require it.
• Abuse monitoring: account, device, IP, payment, upload, moderation, and usage signals may be reviewed to detect repeated abuse or fraud.
• Manual review: user reports, DMCA/IPR notices, severe safety issues, and moderation disputes may be reviewed by Refini staff.

4.1 Prompt and Content Blocklist

Refini maintains a multi-layer prompt and content screening system. Detailed rules are continuously adjusted in response to new evasion techniques and are not publicly disclosed because publishing them could help users bypass moderation.

4.2 Third-Party AI Models and Infrastructure

Refini's image editing and generation features are provided through the following third-party AI models and platforms, all of which are lawfully licensed commercial API services:

• Image restoration, enhancement, and super-resolution: industry-standard commercial models through Replicate, including google/nano-banana, philz1337x/crystal-upscaler, and topazlabs/image-upscale.
• Background and object editing (background removal, replacement and blur; object removal; image extension and outpainting; colorization; and relighting): through the BRIA API.
• AI image generation, sketch-to-image conversion, and multi-image combination: bytedance/seedream-4.5 and openai/gpt-image-2 through Replicate.
• Image content moderation: falcons-ai/nsfw_image_detection through Replicate, with OpenAI omni-moderation-latest in some configured scenarios.
• Text-prompt moderation: local keyword and regular-expression pre-filtering, followed by Meta Llama Guard 4 (meta/llama-guard-4-12b) through Replicate.
• This list may change when models are upgraded or providers change. The latest update date appears in the page footer.

5. Reporting Abuse

Users and third parties can report suspected policy violations, child safety concerns, non-consensual imagery, impersonation, rights violations, or other misuse through Refini's public reporting channel.

• Report Abuse page: https://refini.ai/report-abuse
• Support and safety email: [email protected]
• Copyright and intellectual property claims: https://refini.ai/dmca-policy

6. Review Timelines

Refini triages reports based on severity. Severe child safety, non-consensual intimate imagery, terrorism, credible threats, or clearly illegal content is prioritized for urgent review. Ordinary policy, billing, copyright, or account reports are reviewed as soon as reasonably possible during service hours.

If a report requires more time, Refini may request additional information or preserve relevant evidence while the review continues.

7. Enforcement

When Refini finds a violation, it may take one or more enforcement actions depending on severity, user history, legal duties, and technical feasibility.

• Block or reject a prompt, upload, output, or processing request.
• Remove, restrict, delete, or make inaccessible violating content or related storage objects.
• Limit access to features, credits, subscriptions, downloads, or processing workflows.
• Warn, suspend, or terminate an account for serious or repeated violations.
• Preserve audit logs, hashes, storage keys, moderation evidence, and related records for abuse investigation or legal compliance.
• Escalate severe illegal content to appropriate authorities where required by law.

8. Appeals

If you believe Refini made an incorrect moderation or enforcement decision, contact [email protected] with your account email, relevant image or order identifiers, and a short explanation. Refini will review the available evidence and may restore access where a false positive is confirmed.

9. Policy Updates

Refini may update this policy as laws, platform rules, payment provider requirements, AI model capabilities, and safety risks change. Material changes will be posted on this page and may also be communicated through service notices or email where reasonably required.

10. Contact

General support, safety reports, abuse reports, appeals, legal/privacy requests, billing inquiries, and security reports may be sent to [email protected]. General business inquiries may also be sent to [email protected]. Service hours are 10:00-24:00 UTC+8.